Metaontological Skepticism
Inaugural-Dissertation
zur Erlangung des Doktorgrades
der Philosophie an der Ludwigs-Maximilians-Universität
München
vorgelegt von
Referent: Prof. Godehard Link
Korreferent: Professor Karl-Georg Niebergall
Acknowledgments
This will be short and sweet. For their help by some means or other, I thank Anthony Everett, Andreas Foldenauer, Stasys Hiob, Thomas Hofweber, Herbert Huber, Hannes Leitgeb, Carlos-Ullisses Moulines, Odin Mühlenbein, Jakob Steinbrenner (and, of course, his wife), Brian Weatherson, André Wenzel and Tobias Wilsch.
Special thanks go to Karl-Georg Niebergall, influence and nominalist extraordinaire, and
Alexander Soutschek, who ran the long course, reading late drafts and participating in whole day meetings.
I owe, however, the most and deepest depts to Godehard Link, ∅ystein Linnebo, and Alexander Oldemeier. Especially ∅ystein, and Alex had so much patience and guidance. They were there when work got difficult.
Thanks, to my parents, my two sisters, my grandma and my grandpa, who sadly passed too early to see this, and my uncle Rainer and his wife, my aunt Gundi, for their support and love.
Contents
________________________________________________________
Chapter 1
Introduction
S. 11.1Worrying Questions S. 1 1.2Metaontology S. 2 1.3Skepticism S. 3 1.4Metaontological Skepticism S. 7 1.5Overview S. 9
Chapter 2
Quantifier Variance
S. 112.1 Overview S. 11 2.2 An Attempt to Specify the Theory S. 14 2.3 The Metatheoretical Commitment to
Unique Quantifier Meanings S. 21 2.4 The Collapse of Quantifier Variance in Ordinary
Ontological Positions S. 22 2.5 Conclusion S. 24
Chapter 3
Carnap’s Skepticism
S. 26Chapter 4
The Self-Defeater Argument
S. 524.1 Overview S. 52 4.2 A First Argument S. 54 4.3 A Revealing Response S. 57 4.4 A Better Argument S. 61 4.5 Conclusion S. 63
Chapter 5
Schaffer’s Stroke
S. 645.1 Overview S. 65 5.2 Permissivism and Two Kinds of Existence Questions S. 67 5.2.1 Permissivism and Grounding S. 67 5.2.2 Schaffer’s Diagnosis S. 69 5.3 The Problem S. 69 5.4 A Non-trivial Classical Existence Question Per Se S. 71
5.5 Traditional Existence Questions ≠≠≠≠ Existence
Questions Per Se S. 74
5.5.1 The Epistemological-Descriptivist Case S. 76 5.5.2 The Metaphysical Case S. 80 5.6 Conclusion S. 87
Chapter 6
Direct Reference and Platonism
S. 886.1 Overview S. 88 6.2 Direct Reference, Structured Propositions and Rigid
6.5 Numerals as Quantifiers? S. 115 6.6 Some Related problems S. 117
6.7 Conclusion S. 123
Chapter 7
Ontology and Models
S. 1247.1 Taking Stock S. 125 7.2 Sketches of a New Account S. 126
Appendix:
Triviality
S. 134A.1 Overview S. 134 A .2 Triviality and Philosophy S. 135 A.3 The Role and Irreducibility of Triviality S. 138 A.4 Dilemma S. 141 A.4.1 One Side S. 141 A.4.2 The Other Side S. 142 A.5 The Explication S. 143
A.6 A Formal Model for Triviality S. 147 A.7 Objections S. 151 A.7.1 Context and Triviality S. 151 A.7.2 Triviality and necessary propositions S. 153
1. Introduction
_________________________________________________________________________
1.1 Worrying Questions
How can it be that philosophers haven’t made any progress in ontology for 2500 years? Why do we still not know whether there are abstract objects? Why is it that no rational consensus has been secured over any substantial proposition about what there is?
Metaontological skepticism yields an answer to these questions. The answer is pessimistic. Ontology is not a legitimate philosophical discipline. Therefore it is no wonder that we haven’t made any progress in ontology for such a long time and that no rational consensus has ever been secured over any substantial, foundational proposition about what there is. When we are engaging in ontology, we are engaging in an illegitimate discipline.
But what are the arguments for this claim? Is ontology really an intellectually worthless endeavour?
In this dissertation I will try to answer these questions by outlining, criticizing and assessing the prospects for metaontological skepticism. I will do this by categorizing and laying out the motivations and arguments for various skeptical metaontological positions to show where their strengths and weaknesses lie. Moreover, I will draw some conclusions for disciplines like the philosophy of language, that follow from my work on metaontological skepticism. And I suggest the outlines of an own account of ontology. But before we will get started with this project, I want to flesh out in greater detail, what metaontological skepticism is.
1.2 Metaontology
Metaontology can be characterized as the philosophical study of the foundations, presuppositions and limits of ontology. A study that includes a clarification of the basic concepts and the legitimacy and the cognitive status of ontology, as well as a formulation and investigation of the task and the genuine methods of its subject matter discipline.
You should note however, that the word “metaontology” involves an ambiguity. The ambiguity is of the same kind as the ambiguity in the word “ontology”. Let’s illustrate the ambiguity by means of the – possibly more familiar – word “ontology”. On the one hand, “ontology” denotes a philosophical discipline. Namely, that discipline which is concerned with what there is. The characterization of metaontology sketched out above matches this use of ontology. On the other hand, “ontology” denotes the entities that a theory, a discourse, or a person, assumes or is committed to. For instance, a discourse about whether the electrons under a dielectric in a MOS transistor function as charge carriers assumes that there are electrons under the dielectric of a MOS transistor. A sincere Hawaiian kahuna in the 17th
century reporting his encounters with the ghosts of deceased clansmen assumes that there are ghosts which he can talk to. By the same token, the word “metaontology” can be used to stand for the theory of the nature of ontology which one adopts; it can stand for ones metaontology. In this sense, my metaontology is possibly different from yours. Of course, this cannot be the case, when the first characterization mentioned is adopted.1 In the following, I will use “metaontology” first and foremost in the sense of the first characterization mentioned.
1
You should note that the subject matter of metaphysics cannot be derived from the prefix „meta-“ that obviously. After all, the subject matter of metaphysics is not physics and not any other scientific discipline. Rather, the name “metaphysics” dates back to Andronicus of Rhodes, who edited the Aristotelian works, at around 70 B.C. For the loose body of Aristotelian works which he chose to class after the “physics”, he reserved
the name „ `
τ α µετ α` τ α` ϕυσικ α´ “ which means „the books following the books about the nature”. (It is often said that from a systematic point of view, however, one of the following two suggestions would probably
have been more appropriate: ‘ e
η περι` τ ων~ πρ ωτων´ ϑεωρια´ ’ [„the theory of which is first“;
(Theophrast)], ‘ `
In contemporary philosophy, metaontology enjoys the outstanding status of being – as it is put by the Leeds Gang Ross Cameron and Jason Turner – “the new black”. A 1998 publication by Peter van Inwagen2, bearing the very name “metaontology”, both introduced the name “metaontology” into the jargon of the philosophical community and revived the interest in questions which had already been heavily discussed from the 1940’s to the 1970’s under the lead of Quine and Carnap but somehow had lost its attraction to philosophers during the 1980’s and 1990’s. Some paradigmatic examples for metaontological questions are, “Is being an activity?”, “Is being the same as existence?”, or “Does the existential quantifier adequately capture the single sense of existence?”.
Obviously, Van Inwagen introduced the name “metaontology” to stress its character as a metadiscipline. Metaontology is about ontology. It is a discipline about a discipline. Insofar, it is like metaethics, metaphilosophy, or metamathematics. But - like metaethics and very much unlike metaphilosophy and metamathematics - it is no part of its subject-matter. Metamathematics is obviously still a part of mathematics, metaphilosophy still a part of philosophy.3 So much about metaontology. What about skepticism?
1.3 Skepticism
Philosophical skepticism stirs up most dust in epistemology. Most notable are the skepticisms about knowledge and justification. However, skepticism is not solely restricted to epistemology. There is skepticism about meaning, and reference. About the past, and the future. About the existence of time and so on and so forth. In short, one can be a skeptic about almost everything which can be seriously discussed and thought about. Since so much attention, though, is brought to skepticism in epistemology, the extensive research done in this philosophical area about this topic is very helpful for illustrating what makes up philosophical skepticism in general and metaontological skepticism in particular.4
Skepticism in epistemology would not be very interesting if the subject matter of its doubt was the proposition that we know exactly that the visitors of Munich’s tourist magnet “Wiesn” will have sunny weather on September 28th seven years from now. No one believes
2 Van Inwagen (1998)
3 Cameron (2008), Turner (2008d).
that we know this. No one expects it. In fact, everyone doubts it.5 But skepticism in epistemology goes further. The subject matter of its doubt is not a solitary proposition like the one just mentioned. Rather, it extends to whole kinds of propositions - following Klein (2005) I will call them epistemologically interesting propositions or EI-type propositions - which
contain tokens, many of which are generally thought to be known, given what we ordinarily take knowledge to be. It concerns, for instance, knowledge about the future, the present or the past, knowledge about other people’s minds, or about the “external world.” Everyone believes that we have such knowledge. Everyone expects it and no one doubts it. So if the skeptic was right and we would be wrong this would be a surprising result.
Now, consider some proposition p. Basically, there are only three possible attitudes one can have towards p's truth when considering whether p is true. I say “basically”, since even
though there are - of course - multifarious attitudes one could have towards p all these attitudes towards a proposition can be reduced to the following three basic types of attitudes:
1. One can assent to p. 2. One can dissent to p.
3. One can remain agnostic as to whether p or non-p.
You might be happy or sad that p. Or which is much more common, you might simply be uninterested as to whether p. But these attitudes are parasitic on one of the three basic types. When we are happy or sad that p we are happy or sad that p is true, when we assent to p. Being happy or sad results from our assenting to p or non-p. When we are uninterested in p we are not considering whether p is true and we remain agnostic as to whether p or non-p (at least in some cases).
The proposition, We can have knowledge of epistemologically interesting propositions is
about the very scope of our knowledge. Given what we have said in the last paragraph we can
have the following three attitudes towards this proposition:
(1) One can assent that we can have knowledge of EI-type propositions. (2) One can dissent that we can have knowledge of EI-type propositions.
(3) One can remain agnostic as to whether we can or cannot have knowledge of EI-type propositions.
I will call adherents of (1) Epistemists. Epistemism is the typical non-skeptical position shared by almost all epistemologists today. This position stands in contrast to the skeptical positions expressed by (2) and (3). For the adherents of (2), I will reserve the name Cartesian Skeptics
or simply Cartesians although one can also find different names like Academic Skeptics (and
correspondingly Academics) or possible /switched world skeptics circulating in the literature.6 It is not hard to see that the name “Cartesians” dates back to René Descartes (1596 – 1650) who can be seen as the inventor of modern skepticism in the mid-17th century.7 Whilst the name “Academics” was first used by Sextus Empiricus (around 200 A.C.) to refer to the leaders of the Academy (founded by Plato) during the 3rd to 1st century B.C. Of course it is a controversial question among historians of philosophy whether they really doubted knowledge of EI-type propositions. Switched world skepticism or possible world skepticism typically involve imagining oneself to be in some possible world that is both vastly different from the actual world and at the same time absolutely indistinguishable (at least by us) from the actual world.8 What underlies this form of skepticism is assent to the proposition that we cannot know EI-type propositions because our evidence is inadequate.
Adherents of (3) will be called Pyrrhonian Skeptics after Pyrrho (ca 365 – ca 275 B.C). The
main part of what we know about Pyrrhonian Skepticism is due to the writings of Sextus Empiricus (at around the end of the 2nd century AD). The Pyrrhonians’ aim in remaining agnostic as to whether we can or cannot have knowledge of EI-type propositions, is to being able to withhold assent to all propositions about which genuine epistemological dispute was possible. Thus they did not fall prey to the dogmatism of the Epistemists and Cartesians. According to them, both Epistemists and Cartesians were likewise dogmatic in assenting to the - by pure arguments - unjustifiable and never establishable proposition that we can have
6 Of course, there is this other notorious form of “possible world skepcitism”, namely Quine’s, according to which there are no such things as possible worlds.
7 Cf. Popkin (2003). To term Descartes “the inventor of skepticism” would be a bit unfair. After all, his famous development of skepticism culminating in the famous Cogito-argument is a product of the work of a huge load of other philosophers who had worked on this topic in the two centuries before Descartes. But Descartes was also the one who radicalized skepticism in such an genuinely new, influential and brilliant way that no one afterwards working on skepticism could have failed to learn from his work. Given that, I think it not completely unfair to call him the “inventor of modern skepticism”.
knowledge, or in assenting to the denial of that claim, respectively. The Pyrrhonian accuses the Cartesian Skeptic of holding an inconsistent and self-defeating position by uncritically relying on the ability of reason in his arguments concerning epistemological propositions about EI-type propositions. This problem is famously called the “Cartesian circle” and used to be one of the big problems in Middle Ages philosophy. To be more precise, as you certainly remember, in his First Meditation Descartes, starts out as follows:
“Several years have now elapsed since I first became aware that I had accepted, even from my youth, many false opinions for true, and that consequently what I afterward based on such principles was highly doubtful; and from that time I was convinced of the necessity of undertaking once in my life to rid myself of all the opinions I had adopted, and of commencing anew the work of building from the foundation, if I desired to establish a firm and abiding superstructure in the sciences.” (Descartes 1641: 11).
1.4 Metaontological Skepticism
In metaontology, a corresponding distinction to the just mentioned trichotomy between Epistemists, Cartesians and Pyrrhonians can be applied to describe the various positions toward the question, “Is ontology a legitimate philosophical discipline?”.
First, there are the ones who hold that ontology is a legitimate philosophical discipline. These are all the nominalists and realists, all the ones who seriously engage into ontology. Call them
Ontists. Second, there are the ones who deny that ontology is a legitimate philosophical
discipline, which I will call metaontological Cartesians. And, third, there are the ones who
want to remain agnostic as to whether ontology is a meaningless discipline. These will be called metaontologicalPyrrhonians.
Let’s have a look into which category the different skeptical metaontological theories discussed in this work fall.
The theory of quantifier variance, the first skeptical metaontological theory I will discuss in this dissertation, holds that ontological disputes are nothing over and above trivial disputes about how we should use our language. Insofar ontological disputes are on the same level as disputes about whether we should use “mile” for a nautical mile or for a statute mile. The quantifier variantist even thinks that for any ontological view there is a language such that the ontological view comes out true in this language. Given that quantifier variantists say that ontological sentences are trivial, ontology is not a legitimate discipline. Thus quantifier variantism is a Cartesian metaontological position.
Jonathan Schaffer, one of the most influential (some even say the most influential) contemporary philosopher(s), developed another Cartesian skeptical metaontological account. This is the third skeptical theory that I will discuss in this dissertation. According to his metaontological account traditional metaphysics without ontology is as informative as it is with ontology. Schaffer thinks that the thesis that (almost) everything exists, a thesis, which Schaffer calls permissivism, holds only for some special kind of existence question. Given
that the traditional existence questions (as for instance “Are there numbers?”) are in his opinion of this special kind, permissivism holds for the traditional existence claims. But it does so if and only if ontology does not add any informative content to metaphysics. So, ontology does not add any informative content to metaphysics. Schaffer concludes that ontology is an illegitimate philosophical discipline. Given his permissivism (“everything exists”) Schaffer is a first-order Ontist. But on a higher-order, on a metaontological level he is a genuine Cartesian.
It is important to note that all the mentioned skeptical metaontological positions are motivated by linguistic considerations. They all develop their skepticism out of the assumption that ontological sentences are either trivial, or meaningless, or both, in short, that they are without cognitive significance.
The quantifier variantist argues that ontological disputes are nothing over and above trivial considerations about which language or which conventions in a language to choose. Carnap argues that ontological sentences are trivial within a framework and meaningless outside of such a framework. Schaffer argued that ontological sentences are trivial, since there trivially exists anything.
The overall reasoning of these skeptics can be captured by the following argument.
LINGUISTICALLY-MOTIVATED ONTOLOGICAL SKEPTICISM:
(P1) Ontological sentences are without cognitive significance.
(P2) If the sentences of a supposed philosophical discipline are without cognitive significance, the discipline is illegitimate/valueless.
Even though other forms of metaontological skepticism are conceivable, all currently available skeptical metaontological theories are linguistically-motivated in such a way.
Thus, for the most part of this dissertation, I will be concerned with this kind of metaontological skepticism. Only in the end I will sketch other non-linguistically motivated forms of metaontological skepticism. Note that when I speak of metaontological skepticism in this dissertation, I mean the linguistically-motivated metaontological skepticism, if I don’t say
otherwise. Finally, here’s an overview of the dissertation.
1.5 Overview
In the second chapter of this dissertation, I will be concerned with the theory of quantifier variance. I will argue that the theory of quantifier variance is not a genuinely skeptical metaontological position. Rather, quantifier variance is an ontic metaontological position in disguise. Furthermore, I show that the quantifier variantist has problems in accounting the tight connection between language and beliefs.
The third chapter contains a discussion of Carnap’s sceptical metaontological account. I will show that a proper understanding of Carnap’s work from 1950 cannot be gained without a prior understanding of its intellectual roots in the philosophy of the logical positivists, the Vienna Circle and other writings of Carnap’s from the post-Vienna period. On this base I will give a thorough reconstruction of Carnap’s 1950 account and the involved notion of a
linguistic framework. Then I will rebut the currently widespread move to identify Carnap’s
theory with the theory of quantifier variance.
In the fourth chapter, I give an argument against Carnap’s account in particular and against some forms of linguistically-motivated metaontological skepticism in general. I will argue against these forms of metaontological skepticism. They are self-defeating. Even though metaontological skepticism is not self-defeater instable per se, the most promising and widely
adopted ways of developing it are.
question, existence questions per se. In these arguments, I adopt methods and theories of contemporary philosophy of language. This paves the way for the discussion in the next chapter.
In that chapter, the sixth chapter, I show that our ignorance about what numbers are, leads to a hard problem for the standard theory of reference, the theory of direct reference. Direct reference theorists hold that the meaning of singular terms is completely exhausted by their referents. In the case of numerals then, the meaning of numerals is completely exhausted by the numbers they refer to. I will argue that this assumption leads to severe problems in accounting for how we can understand mathematical sentences.
The seventh chapter contains some concluding remarks, but also the outlines of a new approach. Even though ontology may not be able to tell us what exists, this does not render ontology illegitimate or valueless. Rather, the (or a) task of ontology is to build good models that can serve as frames for the theories of other philosophical or non-philosophical disciplines.
2. Quantifier Variance
_______________________________________________________
This chapter contains my first discussion of a position commonly running under the label “skeptical metaontological”. This position is the theory of quantifier variance. The theory of quantifier variance is explicitly defended in Hirsch (2002), (2004) and Turner (2008a) and (2008b) but often also ascribed to Putnam (1987a) and (1987b) and Rudolf Carnap (1950). In recent years, the theory of quantifier variance has gained a lot of attention in metaontology. Examples are Sider (2001), (2004), (2006), (2007) and (2008), Hawthorne (2006), and (2008), Eklund (2007), (2008a), and (2008b), Bennett (2008), Chalmers (2008), Hale and Wright (2008). I will first state the theory, both in an informal and a formal way, and, secondly work out three arguments against the theory of quantifier variance. The upshot will be that the theory of quantifier variance is not a genuinely skeptical metaontological position.
2.1 Overview
For Hirsch ontological disputes are nothing over and above trivial disputes about how we should use certain expressions. Hirsch holds that for any ontological position a language exists whose ontological vocabulary (as for instance the existential quantifier) possesses such a meaning that the claims of this ontological position come out true in this language.
Ontological dispute is as trivial as the dispute about whether one should use the expression “mile” synonymous with the expression “nautical mile” or “statute mile.”
As an example, consider the following situation. A British and a French sailor talk about whether one should use the car to get from village A to village B or whether one can also use the bike. It is common knowledge among the two conversational partners that the distance between the two villages amounts to 1852 metres.
What do the two sailors argue about? They certainly do not argue about whether the distance between A and B amounts to 1852m. This is common knowledge among them. Rather, they only argue about whether “mile” should be used synonymous with “nautical mile” or with “statute mile”. While the French sailor argues that “mile” should be used with “nautical mile,” the British sailor heavily disagrees. For him the word “mile” should be reserved as being synonymous with “statute mile”. Thus they only argue about how words should be used and about which conventions for a word should be adopted. In other words, their dispute was
merely verbal. Should we use the convention that “mile” denotes a distance of 1852m or that
a mile denotes a distance of about 1609m? Our normal reaction would be to claim that we couldn’t care less which expression to use. After all it is pretty pointless to say that one of the options fares generally better than the other. They are on a par.
According to Hirsch, this is exactly what happens when two ontologists argue about what exists. He writes:
“What the doctrine does imply is that our linguistic decisions determine the meaning of the expression “there exists something”; hence, they determine the meaning of the sentence “There exists something composed of Clinton’s nose and the Eiffel Tower”. Hence, the truth or falsity of this sentence depends in part on our linguistic decisions. It is merely a use-mention confusion to conclude that whether or not there exists something composed of Clinton and the Eiffel Tower depends on our linguistic decisions” (Hirsch 2002: 52).
are no abstract objects” comes out true, while in the platonist’s language, “there are abstract objects” comes out true.
Another example is the debate over the ontology of material objects. This debate centers around the question “When are given material objects part of some further composite object?”. A mereological universalist, Lewis (1991) or Link (1998) for instance, wants to say that given material objects always form some further composite object. Thus, Kurt Gödel’s pen and Alfred Tarski’s wife form a composite object. Others, like Dorr (2005), say that this is never the case. There are no composite material objects. Some, like Van Inwagen (1987), say that composite objects sometimes, but not always, exist. Only if objects stand in the appropriate relation to each other like being appropriately glued together, there exists a further material object that they compose. Here Hirsch thinks, that the different factions do not argue about whether reality is made up by composite material objects, but rather about which meanings we should give to our words. Their dispute is again merely verbal. They only argue about whether the ontological vocabulary should be used as in the mereological universalist’s, the mereological nihilist’s or the mereological moderatist’s language. All languages are on a par. No language is intrinsically better than any other. Hirsch writes:
“There are many possible perspectives on ‘the existence of objects’, which are all adequate for describing the same facts, the ‘same way the world is’” (2004: 231).
So, according to Hirsch, a quantifier variantist will “…address a typical question of ontology either by shrugging it off with Carnapian tolerance for many different answers, or by insisting with Austanian glee that the answer is laughably trivial” (2002: 67).
To put it in a nutshell, the theory of quantifier variance is the conjunction of the following two theses:
First. For each ontological position, there is a language, such that the quantificational expressions can be interpreted such that the theory comes out true.
Second. None of these interpretations and languages is any “better” than the others.
express other things than the second-order property of being non-empty? And why should there be any such languages or interpretations? We do yet not have answers to these questions. We need a sharper formulation of the theory of quantifier variance.
2.2 An Attempt to Specify the Theory
9An explication of the theory of quantifier variance must satisfy firstly an appropriate account of what these quantifier meanings are and secondly of what is understood under “better”. There are two natural suggestions to answer the first question. These suggestions are based on the fact that a quantifier Q is normally understood as ascribing to its domain D a property P. Thus changing the meaning of the quantifier Q either involves changing domain D or property P.
The first natural suggestion (made by ∅ystein Linnebo in p.c.) to answer the first question does not work. The suggestion is that the quantifier meanings are just the restrictions of the quantifiers. This idea has it that the mereological universalist and the mereological nihilist only differ by restricting their quantifiers in unequal ways. The mereological nihilist’s quantifier for example is restricted to all atoms, while the ontological pluralist’s quantifier is not restricted to atoms only, but also quantifies over the composite material objects made up by these atoms. The problem with this suggestion is that if a quantifier is restricted, one just ignores some of the things which one excludes by the restriction. But the things are still there. According to this suggestion, the mereological nihilist would never be a real nihilist but rather a mereological universalist who just ignores for a certain while some of the things a universalist believes in.
The second suggestion to answer the first question does not work either. Normally the existential quantifier is understood as expressing that its domain is non-empty. Or in other words, the existential quantifier expresses the second-order property of sets of being non-empty or, again in slightly different terms, the second-order property of properties of being instantiated at least once. This is the overlap property of the existential quantifier. So, if the
meanings of the quantifiers vary with the languages, one could argue that the existential quantifiers in these languages express other second-order properties like the property of being
instantiated at least twice (or being at least two-membered), or being instantiated at least three times (or being at least three membered), etc. However, this suggestion doesn’t hold, since the inference rules for the existential quantifier would have to be changed accordingly in the different languages. But the quantifier variantist does not want to change the inference rules. Thus, the second suggestion can not be what the quantifier variantist means.
This leaves it pretty mysterious what the quantifier variantist’s claim about the different meanings of the existential quantifiers boils down to. In fact, I think that the theory of quantifier variance is already dead at this point. But let’s see whether we can make allowances to the quantifier variantist. In face of the quantifier variantist’s enigmatic quantifier meanings, the most reasonable way out of this lacuna is giving up the task of trying to understand what these quantifier meanings are supposed to be. Rather, I will follow the
lead of Sider (2007), who has done as much as no other to give a better formulation of the theory of quantifier, in only trying to understand what the quantifier meanings are supposed to
do.
Let’s turn back to the existential quantifier. If one accepts the results of Kripke’s thought experiment, there might be uncountable different meanings for the existential quantifier. If one accepts Lewis’s general outlook, these different meanings for the quantifier might differ from each other in their degree of naturalness as meanings of the quantifier. As Sider (2003: 144) says:
“existence is the one and only highly eligible meaning that fits our use of (unrestricted) quantificational expressions”.
For the sake of simplicity, I will join Sider in fleshing out the notion of “being a better meaning of a quantifier” with the notion of “being a more natural meaning (in the sense just discussed) of a quantifier”. However, I don’t think that one is forced to do that. Other ways are always open, as an axiomatic treatment of the “better than”-relation.
Above, we concluded that it is wholly mysterious what the quantifier meanings are. Therefore, we gave up the attempt to say what they are and confined us to the task of saying what they are supposed to do. Three demands have to be met by quantifier meanings. First it must be possible to ascribe the notion of naturalness to them. Second they have to play a part in determining the truth-values for quantified sentences. Third it must be possible to say that there are more or less “expansive” quantifier meanings, where this is not the same as varying domain restrictions.
Sider (2007: 11) satisfies these three demands by introducing three undefined relations for a meaning-context pair <m,c>. (Contexts are necessary as an explanation of contextual variation of quantifier domains). The relations are as follows.
(1) Meaning m is at least as natural as meaning m’.
(2) Model M depicts meaning-context pair <m,c>.
(3) Context c belongs to meaning m.10
10 Sider doesn’t define “meaning”, “naturalness”, and so on. One could do this by giving axioms which govern the relations. So, for instance, the relation “at least as natural as” could be governed by the following axioms where the quantifiers range over meanings, where “N” represents the relation “is at least as natural as”:
The first relation is supposed to help answer the first constraint, the second one the second constraint, and the third one the third constraint. If Q is a meaning context pair <m,c>, such that (2) is satisfied, then Q is called a quantifier.
By means of clause (2), a definition of truth for a sentence relative to a given meaning can be given. The definition is as follows:
DEFINITION 1 Sentence φ is truecm if and only if φ is true in some model that depicts <m,c>.
Moreover, let the following be true:
(a) For each context, there is exactly one meaning to which it belongs. (b) No model depicts anything other than a quantifier.
(c) Each quantifier is depicted by some model.
(d) The same sentences are true in any two models that depict the same quantifier.
So far we have shown how the first two constraints can be satisfied. But what about the third constraint? According to the third constraint, it must be possible to say that there are more or less “expansive” quantifier meanings, where this is not identical with varying domain restrictions. So how can we model these expansions, that are independent of domain restrictions? A possibility is to adopt the following definition:
(N3) ¬∀x∀y (Nxy∧Nyx→x=y)
With the relation “is at least as natural as”, it is simple to define the relation “is as natural as” thus: N=xy :↔ Nxy∧Nyx,
and the relation “is more natural than” thus: N+xy :↔ Nxy∧¬ N=xy.
Antisymmetry, which is the negation of (N3) doesn’t hold. That is, it doesn’t hold that:
DEFINITION 2 Quantifer Q expands quantifier Q’ if and only if every model that depicts Q’ has a supermodel that depicts Q. Q properlyexpands Q’ if and only if in addition, Q’does not expand Q.
This definition captures the idea that a mere restriction changes the context but retains the same meaning, whereas the distinctive kind of expansion changes the meaning as well as the context. This is achieved as follows: We can expand a quantifier in another way than by manipulating its domain that is blowing up or downsizing it. Proper expansion collapses into ordinary restriction, if <m,c> properly expands <m’,c’> and m=m’, because then only the contexts c and c’ differ. After all, the restrictions on the domains are dependent on contexts. A quantifier determines together with a context a restriction of the domain.
But here we have the possibility of some other kind of expansion, if m≠m’.
Now we are able to state the first versions of the thesis of the quantifier variantist. Let M be a class of models. The models in M are regarded as “quantifier worlds”. This is intended to capture the idea that the models describe what the world would be like given various quantifier meanings. After all, as Harold Hodes (1991a) once elegantly put it: Truth in a model is a model of truth. If E is a set of meanings, let Q(E)={<m,c> | c belongs to m and m∈E} be the set of quantifiers that are grounded in E. Q(E) is made up by the multiple candidate quantifier meanings in which the quantifier variantist believes. Any such claim will say roughly that each quantifier world in M depicts some quantifier in Q(E).
How do quantifier worlds and quantifiers correspond to each other? Here are four suggestions:
THESIS 1 Weak M/E-quantifier variance
Every member of M depicts some member of Q(E)
But Weak M/E-quantifier variance allows that all the quantifiers in Q(E) are restrictions on a single maximal quantifier.
THESIS 2 Moderate M/E-quantifier variance
Weak M/E-quantifier variance + some member of M outruns some member of E,
where model M outruns meaning m if and only if for no c does M depict <m,c>. This version avoids the pitfalls of THESIS 1 by prohibiting that every quantifier world is covered by only one meaning.
THESIS 3 Strong M/E-quantifier variance
Weak M/E-quantifier variance + every M∈M outruns some member of E (provided M is a proper extension of some member of M).
THESIS 4 Unrestricted M/E-quantifier variance
Every member of M depicts some unrestricted member of Q(E).
Strong quantifier variance goes further than both THESIS 1 and THESIS 2 by claiming that each quantifier world is beyond the reach of some meaning (except when the world is a “minimal” member of M). Unrestricted quantifier variance goes even further than THESIS 3 by claiming that each quantifier world depicts some unrestricted (!) quantifier.
What is the content of these four theses? What is the range of quantifier worlds?
UPWARD CLOSURE
Any supermodel of a member of M is itself a member of M.
The next definition is as follows:
DOWNWARD E-CLOSURE
If M is a supermodel of M’ and M depicts some member of Q(E), then M’ outruns some member of E.
and we choose a reduct of that model, then there is some other language we can speak in which we will not reach the chosen reduct. Moreover, reaching this reduct is independent of quantifier restriction. All the same, Sider (2007: 15) thinks that the quantifier variantist may perhaps not want downward closure. After all, some sentences could be atomic, in that no meaning treats them as false except because of quantifier restriction. Sider (2007: 15) thinks that the sentence ‘There exist electrons’ might be an example. Atomicity can be defined as follows:
DEFINITION 3 Sentence φ is E-atomic if and only if for every m∈E, there exists a c such that i) φ is truecm, and ii) for any c’, if φ is not truec’m then <m,c’> is a restriction of <m,c>.
I doubt that the quantifier variantist really wants to avoid this claim. An idealist could at last reject the sentence “There exist electrons”. In ontology there are no atomic sentences in the sense that every ontologist would accept them. If there were, some of the most important ontological positions could be ruled out a priori.
Nonlogical expressions can undergo expansion, in extensions of models either if new non-logical expressions not interpreted in the reduct are interpreted in the extensions of the models, or if the extensions of the non-logical expressions of the reduct are enlarged in the extensions of these models. This is captured in the following definition.
DEFINITION 4 M is a (proper) <K,L>-supermodel of M’ if and only if i) M is a (proper) supermodel of M’, ii) any nonlogical expressions that are newly interpreted (i.e., interpreted
by M but not by M’) are in K, and iii) any nonlogical expressions that are altered (i.e., have different extensions in M and M’) are in L.
mereological universalists and nihilists. NeoCarnapian quantifier variance claims may therefore have the form <∅,{‘part of’, ‘overlaps’, . . . }>, where the empty set as the first member of the pair indicates that no new non-logical expressions were introduced.
In the face of all this, we could state the theory of quantifier variance, for instance, as follows:
NeoCarnapian quantifier variance There is a nonempty class of models, M, and a class of meanings, E, such that:
i) M obeys upward <∅,{‘part of’, ‘material object’ . . . }> - closure
ii) every member of E is as natural as every other, and no meaning not in E is as natural as any meaning in E
iii) strong M/E-<∅,{‘part of’, ‘material object’, . . . }>-quantifier variance is true.
With these formal remarks in mind, we can turn to the next part of the chapter. This part contains some criticisms of the theory of quantifier variance.
2.3 The Metatheoretical Commitment to Unique Quantifier Meanings
As we have seen, we can informally characterize a quantifier variantist as someone who believes that there are multiple meanings for the existential quantifier such that all these meanings are on a par. No one of these meanings is somehow better or worse than the others. All are equally natural meanings. In our strict formal definition of quantifier variance this informal idea was characterized as follows:
NeoCarnapian quantifier variance There is a nonempty class of models,
M, and a class of meanings, E, such that:
i) M obeys upward <∅,{‘part of’, ‘material object’ . . . }> - closure
ii) every member of E is as natural as every other, and no meaning not in E is as natural as any meaning in E
But these two characterizations pose a devastating threat for the quantifier variantist: a self-defeater is lurking. If the quantifier variantist is someone who thinks that there are multiple meanings for the existential quantifier, then he uses an existential quantifier in the formulation of his claim. In the formal definition he is committed to the claim that there is both a
nonempty class of models and a class of meanings. But if he uses an existential quantifier in his claim, his claim is susceptible to his own thesis. This implies that there are numerous ways of how to interpret his thesis which does not express something substantially about the world, but only about which conventions we should accept, to govern the use of our language. Quantifier variantism becomes a triviality. But this is not what the quantifier variantist wants. He wants to make a substantial claim about ontology.
It is hard to see how the quantifier variantist can counter this objection. After all, it is he, who introduced the thesis. Maybe he could find a paraphrase which avoids the commitment to abstract entities. However, I doubt that there is such a paraphrase which leaves the language-dependence of the quantifier theorist’s thesis as strong as in the above (formal) definition. The onus of proof is on the quantifier variantist’s side.
2.4 The Collapse of Quantifier Variance in Ordinary Ontological Positions
As we have seen, the quantifier variantist believes that ontological debates are shallow. When two ontologists argue about whether there are Fs, they are not engaging in any genuine ontological debate. They are not arguing about what stuff reality is made up of. They are only talking past each other. In fact, they argue about which linguistic conventions we should use. Thus, their disagreement is merely verbal. They are not different from the two sailors from above arguing as to whether 1852m make up a mile or not. The two sailors are just arguing about whether one should use “mile” as synonymous with “nautical mile” or “statute mile”.
Likewise, assume that we have answered the question about which ontological language we should use. Nominalists and Platonists, Mereological Universalists and Nihilists have all agreed upon which conventions they should use to govern the use of our ontological vocabulary. What happens if this “better stadium of the philosophical world” has arrived? According to the quantifier variantist, it will turn out, that all ontological questions can be settled trivially. If we have convened upon which language to use, we have agreed upon which language to use. Ontological sentences are still true in such a case. But this seems to mean that ontological questions can still be legitimately raised. There still exists the question as to whether there are Fs or not. So far the quantifier variantist.
However, the analogy which the quantifier variantist persues does not carry over to the ontology case. The dispute between the two sailors, as to whether 1852m constitute a mile is certainly trivial, when they have settled upon how to use the word “mile”. This is due to the fact, that it is trivial for the first sailor that 1852m is a mile and trivial to the second sailor that 1852m is not a mile. After they have settled their use, they just agree on one way of understanding “mile”. This doesn’t lead to a change in triviality since it has been a trivial matter on both counts. But, it is not trivial, whether there are abstract entities, even when one has settled upon how to use his ontological vocabulary. After all, philosophers need hard arguments to establish the claim that there are abstract objects or that there are no abstract objects. Almost no one thinks that it is trivial whether there are abstract objects or not.11 We don’t know whether there are abstract objects or not. And we need arguments to dispel our ignorance. But this means that there is also no triviality which can carry over to the question as to whether there are abstract entities. The analogy between “mile” and ontological vocabulary has reached its limits. Thus, the quantifier variantist is clearly in need of an account which explains why ontological disputes are nevertheless trivial. The onus of proof lies on his side. Therefore, it cannot be that ontological disputes are trivial. Therefore, quantifier realism is not a genuinely skeptical metaontological position.
The theory of quantifier variance furthermore poses some urging epistemological problems. The quantifier variantist does not make any normative claim about how ontology should be done. Rather, quantifier variantism is a descriptive metatheory of ontology. The quantifier variantist claims that his thesis provides an adequate description of what happens when people engage in the philosophical business of metaontology. Still almost no ontologist thinks that
the quantifier variantist is right and that he provides a correct description of what he does. Why is this?
The quantifier variantist holds a view according to which our ignorance about our own language is extreme. The quantifier variantist maintains that we do not know anything about the meanings of the ontological vocabulary of our languages. Given that we have to use ontological vocabulary all the time in daily life, our semantic competence is minimal.
But now consider an ontologist called Quiner, who starts out his philosophical career as a
nominalist and ends it as a platonist. Normally one would say that the positions he defended over the course of his life are contradictory. First he defended a view according to which there are no abstract entities, then he defended a view according to which there are abstract entities. But according to the quantifier variantist, such a philosopher just changed the languages in which he speaks over the course of his life. The quantifier variantist says that he did not realize that he changed the languages in which he spoke. In fact, he thinks that he still speaks the same language. So the positions he defended over the course of his life are not contradictory. So far the quantifier theorist has no problems in explaining what is going on in such a case. But didn’t Quiner also preserve his beliefs? The answer is: no. In the early stages of his career Quiner believes that there are no numbers. In the later stages of his career Quiner believes that there are numbers. This is due to Quiner’s extreme ignorance about the meanings of the ontological vocabulary: After all, he doesn’t know that the positions he defends are not contradictory. Nevertheless, the beliefs Quiner held over the course of his career are contradictory. This makes it very hard to see how the tight connection between our language and our thought can be preserved by the quantifier variantist. Moreover, it makes it very hard to explain, how communication should be possible. If we do change our beliefs over time, but cannot express this change in language, how can we ever confer our beliefs to someone else? Given this, I reject the theory of quantifier variance.
2.5 Conclusion
In this chapter, I first tried to motivate the theory of quantifier variance and stated it both informally and formally. The formal statement of the theory took place in an algebraic setting. The advantage of this setting was that we were not forced to specify what the prima facie
3. Carnap’s Skepticism
_________________________________________________________________________
The first attempts to cast the ship of logic off from the terra firma of the classical forms were certainly bold ones, considered from the historical
point of view. But they were hampered by the striving after ‘Correctness.’
Now, however, that impediment has been overcome, and before us lies the
boundless ocean of unlimitied possibilities.
Carnap(1934), The Logical Syntax of Language
The subject matter of this chapter is the skeptical metaontological position developed in Carnap (1950).12 This account is generally conceived of as the father of all the skeptical metaontological accounts in the 20th century.
3.1 Overview
The skeptical metaontological account which Carnap developed in his (1950) is the most influential skeptical metaontological account in contemporary philosophy. There is no introductory class in metaphysics which fails to discuss this seminal piece of work and no discussion of metaontological skepticism which misses to mention it.
In the following chapter I will discuss this account. I will show that the fruits of this latter opus can already be seen in its roots in earlier work; a proper understanding of the account can only be achieved by reviewing its intellectual roots in the philosophy of the logical positivists, the Vienna Circle and other writings of Carnap’s from the post-Vienna period. This will be done in sections 2 – 4. Sections 5 and 6 contain a presentation of Carnap’s 1950 account in general and his notion of linguistic framework in particular.
The work done in sections 5 and 6 will form the basis for my rebuttal of a widespread misconception of Carnap’s position in contemporary metaontology. This misconception is the tendency to assimilate Carnap’s metaontological skepticism to the theory of quantifier variance which was discussed in the former chapter. The rebuttal will take place in section 7. Section 8 sums up.
3.2 The History of Carnap’s Skepticism
In his (1950), Rudolf Carnap famously argued that the existential statements which philosophers are interested in (as e.g. “are there abstract objects?”) are meaningless. This skepticism towards ontological matters does not come out of the blue. Rather, it can be traced back to a more comprehensive skepticism towards metaphysics in general. Carnap shared this attitude towards metaphysics with the other logical positivists of the Vienna Circle.13 As is well known, the Vienna Circle was a group of German and Austrian philosophers who regularly met to discuss various philosophical problems in Vienna of the 1920’s and early 1930’s. Basically, the group included Otto Neurath, Rudolf Carnap, Hans Hahn, Moritz Schlick, Kurt Gödel (at least partly), Friedrich Waismann, Karl Menger, Philipp Frank, Gustav Bergman, and Edgar Zilsel.
The skeptical metaontological outlook of the Vienna Circle is one of the few plump pillars logical positivism is often reduced to.14 Other pillars are the verification principle of meaning and a naïve reductionism which was allegedly defended by the members of the Vienna
13
An exception is, of course, Gödel, who defended a radical version of platonism. To some readers, reckoning Gödel to the Circle may seem dubitable. After all, he didn’t attend every meeting and didn’t share many of the logical positivists’ beliefs. I agree with this. There is another sense in which the reckoning is pretty uncontroversial. The Vienna Circle can be regarded as a mutual intellectual influence community. Gödel had a big influence on the other members of Circle as had the other members of the Circle a big influence on Gödel (cf. Goldfarb (2005)) . In his (1934), Carnap, for instance, thanks Gödel in the Preface for having read a draft of the book and already astonishingly exploits the Gödelian method of arithmetization in the book. (Carnap reports in his (1963b: 53): “In August 1930 he [Gödel] explained to me his new method of correlating numbers with signs and expressions. Thus a theory of the forms of expressions could be formulated with the help of the concepts of arithmetic.”) It is in this latter sense when I reckon Gödel to the members of the Circle and not in the former.
Circle.15 However, it is important to note that the metaontological skepticism of the logical positivists was very sophisticated and complex and cannot be transferred easily to contemporary metametaphysical skepticism.
In fact, the philosophical bogeyman as which logical positivism is willingly presented in introductory classes, is seldom substituted for the serious and sophisticated philosophical position that it has in reality been. The complexity of the demands that the logical positivitst had to satisfy haven’t often been recognized. Many people think what Michael Scrivens (1969: 195) once polemically wrote:
“The Vienna Circle or Wiener Kreis was a band of cutthroats that went
after the fat burghers of Continental metaphysics who had become intolerably inbred and pompously verbose.”
Fundamentally different influences generated the tensions that logical positivism had and wanted to master. What were these influences?
First. The philosophy in Germany from 1910 – 1925 was strongly shaped by the reception, interpretation and further development of Kant’s philosophy. Various different neo-Kantian schools set out to give different solutions to problems of the Kantian philosophy. The two most influential schools were the so-called Marburg School and the Southwest German School. These two schools stood in direct rivalry. Carnap, for instance, was educated by Bruno Bauch, a then famous adherent of the Marburg School. The leading question which was of interest to the logical positivists was whether there was any apriori knowledge.16
Second. The Logical Positivists were strongly influenced by the German mathematical
15
Michael Friedman correctly dismisses the idea of reductionism: “The special sciences – more specifically, the “exact sciences” – are simply taken for granted as paradigmatic of knowledge and certainty. Far from being in a position somehow to justify these sciences from some higher vantage point, it is rather philosophy itself that is inevitably in question. Philosophy, that is, must follow the evolution of the special sciences so as to test itself and, it need be, to reorient itself with respect to the far more certain and secure results of these sciences” (Friedman 1991: 508).
tradition17 and the German tradition in the natural sciences. Schlick, for instance, graduated under Max Planck and was befriend to Einstein. Beside these personal acquaintances there were huge influences from the revolutionary developments in the sciences. To give some examples: Einstein’s theory of relativity, which made it obvious that Kant’s analysis of space and time couldn’t be true (Carnap tried to reconcile the Kantian theory of space and time with the consequences of Einsteinian theory in his dissertation “Der Raum” from 1920), discoveries which led to the development of quantum theory18 and Gödel’s incompleteness theorems.
Third. The positivists were strongly influenced by Russell, Frege and Wittgenstein.19 As you certainly know, Carnap was a student of Frege. Frege’s influence was especially strong in the positivist’s philosophy of language. His conception of analytic truth showed the positivists a way of how to retain a radical empiricism in the spirit of Mill but at the same time avoiding the weakness of traditional empiricism in being able to give a faithful account of mathematics. (As you may remember Frege argued against the Kantian assumption that mathematics is synthetic a priori. Frege tried to show that arithmetic is independent on our spatiotemporal intuition but is built into the general conditions of thought itself.) Even in his later work (1947), Carnap retained the Fregean idea of assigning different semantic dimensions to expressions. All members were acquainted with the writings of Russell on sense data and his and Withehead’s Principia Mathematica and tried to extend the realm of what is mathematically tractable to philosophy. Carnap, for instance, employed this device in the
17
Friedman (1991: 510) thinks that the influence of some of German mathematics was the main impetus on the logical positivists: „The initial impetus for their [the Logical Positivists; WS] philosophizing came rather from late nineteenth-century work on the foundations of geometry by Riemann, Helmholtz, Lie, Klein, and Hilbert …“
18 Hahn (1933: 248 – 49) has the following story: “Vor vielen Jahren, auf einem mit einem Freunde unternommenen Spaziergange im Walde, machten wir, indem wir dem Treiben in einem Ameisenhaufen zusahen, die scherzhafte Bemerkung, die Zoologie könne doch gar nicht davon sprechen, wie sich Ameisen verhalten, sie könne nur davon sprechen, wie sich Ameisen verhalten, wenn Menschen ihnen zusehen; das war ein Scherz, aber es liegt viel Ernst in diesem Scherze: jeder Vorgang wird irgendwie dadurch gestört, dass man ihn beobachtet; die Physik aber spricht vom ungestörten Vorgange – dass das nicht eine zu vernachlässingende Spitzfindigkeit, sondern von prinzipieller Bedeutung ist, wird durch die neueste Entwicklung in der Physik in klares Licht gerückt.“
theory in his Aufbau. As you certainly know, too, there were some meetings with Wittgenstein, parts of which were protocolled by Friedrich Waismann. Wittgenstein inspired the positivists to treat the mathematical sentences as tautologies and a priori.
Fourth. There were social and political aspects. The logical positivists were Marxists (Neurath was even a minister of economics in the short-lived Bavarian Räteregierung under Ebert).
Beside that, they felt a tight connection to the movement of the Neue Sachlichkeit and the Bauhaus in the arts. With this influence they were direct opponents to national-conservative
philosophers like Martin Heidegger.20
A nice quote by Michael Friedman sums up this situation:
“In the European context of the 1920’s, logical positivism arose and developed as a powerful revolutionary force, deeply intertwined with the other revolutionary trends (in the sciences, in the arts, in politics, and in society) that made up what we now know as Weimar culture. The logical positivists aimed at nothing less than a total refashioning of philosophy as a whole, that would definitely end the fruitless, and endless, controversies of traditional metaphysics on behalf of a new “scientific” enterprise in which continuous and cooperative progress could be made solving fundamentally technical problems. And they took their inspiration and their models for such a radical disciplinary refashioning from the breathtaking revolutionary developments simultaneously taking place in mathematical physics and the foundations of mathematics. Although the positivists were, of course, also well aware that there were powerful opposing forces, particularly within German philosophy, working in a quite contrary direction, these developments in the sciences themselves still inspired them, in the words of Carnap’s Aufbau, in “the faith that this
[scientific-philosophical] orientation belongs to the future” (Friedman 1999: xiii).
The skeptical metaphysical outlook of the positivists has its place in this attempt to satisfy all of the above influences. Again Friedman aptly writes:
“It is by no means surprising, therefore, that the logical positivist movement was very actively engaged with the other vocal philosophical movements of the time as well – with neo-Kantianism, with Husserlian phenomenology, even with the “existential-hermeneutical” variant of phenomenology then being initiated by Martin Heidegger. For all of these movements took it upon themselves to venture a radical reform of the German philosophy in which it would renew and reinvigorate itself in a “scientific” spirit, much as the sciences themselves had recently done” (Friedman 1999: xi – xii).
The logical positivists wanted to establish what they called scientific philosophy. Carnap, for instance, writes:
“In our “Vienna Circle”, as well as in kindred groups (in Poland, France, England, USA, and, amongst individuals, even in Germany) the conviction has grown, and is steadily increasing, that metaphysics can make no claim to possessing a scientific character. That part of the work of philosophers which may be held to be scientific in its nature – excluding the empiricial questions which can be referred to empirical science - consists of logical analysis. The aim of logical syntax is to provide a system of concepts, a language, by the help of which the results of logical analysis will be exactly formulable. Philosophy is to be replaced by the logic of science – that is to say, by the logical analysis of the
concepts and sentences of the sciences, for the logical of science is nothing other than the logical syntax of the language of science” (Carnap
1934: xiii).
They wanted to do philosophy in a way which was not inferior to methods which are as precise and formal as in the sciences. Basically, they wanted to achieve this goal by the then newly developed formal devices of Frege and Russell. But still they saw themselves in a tradition of the synthetic a priori inaugurated by Kant. For the positivists metaphysics was a paradigmatic example which violated these demands.
“Dieser Begriff [metaphysics; WS] wird in dieser Arbeit wie in Europa üblich als Bezeichnung für den Bereich angeblichen Wissens über das Wesen der Dinge gebraucht, der sich der empirisch begründeten induktiven Wissenschaft entzieht. Metaphysik in diesem Sinne umfasst Systeme wie die von Fichte, Schelling, Hegel, Bergson, und Heidegger, jedoch nicht Ansätze, die auf eine Synthese und Verallgemeinerung der Ergebnisse der verschiedenen Wissenschaften zielen“ (Carnap 1932: 108).
So what then are the logical positivist’s arguments against metaphysics?
3.3 Examples of Skeptical Metametaphysics
The main direction of impact of the logical positivists’ attack on metaphysics underwent a change over the years. The epistemologically oriented criticism from the first years of the Vienna Circle turned towards the logico-linguistic one of the late years.
Moritz Schlick, for instance, argues in his (1926) that there is no legitimate conception of how information can be gained in metaphysics.21 The suggestions are either entirely meaningless, or they redound to metaphysics collapsing into science.
In particular, Schlick argues as follows. There is a fundamental difference between phenomenal experience (Erleben) and knowledge (Erkenntnis). Knowledge can be gained by
induction. Knowledge is propositional. Propositional structure is expressible by sentences. Sentences have logical forms and stand in logical relations to other sentences. So knowledge is directed at “pure forms” 22 and logical relations. Not so phenomenal experience. Phenomenal experience is neither directed at logical forms nor at logical relations but at “content”.23 It is the humanities and the arts whose first and foremost task it is to animate this experience. Animate and not express it; experience is outright inexpressible. Knowledge transcends experience.
21
See also his (1934)
22 „Erkenntnis ist also ihrem Wesen nach Erkenntnis von Formen, Beziehungen, und nichts anderes“ (op. cit.: 176).
